ICI Atmospheric Profiles Over Rondonia Using HSB Data
Emulated From AIRS Information
Rodrigo Augusto F. de Souza 1, Juan Carlos Ceballos 1, João Carlos Carvalho 2
1
Centro de Previsão de Tempo e Estudos Climáticos / Instituto Nacional de Pesquisas Espaciais
(CPTEC/INPE), Cachoeira Paulista, São Paulo, Brazil. rodrigo@cptec.inpe.br
2
National Oceanic and Atmospheric Administration / National Environmental Satellite, Data, &
Information Service (NOAA/NESDIS), Washington-DC, USA.
Abstract
It was analyzed the performance of the Inversion Coupled with Imager (ICI) software over the
Amazon region, with set up for AMSU/NOAA-16 channels but using information from AQUA
sounding system (AMSU and AIRS). AIRS channels were used to recovering the absence of
HSB data, whose behavior should be similar to AMSU-B. The HSB brightness temperature in
clear sky conditions was simulated by a linear combination of AIRS channels sensitive to water
vapor band. The selected AIRS channels were the five best correlated with HSB (183 GHz) in a
set of 3000 pixels distributed over oceanic areas for one training day (August 31, 2002). The
simulated HSB data together with AMSU-A/AQUA information were used to recover vertical
profiles of temperature and moisture, using ICI with two data bases: a) TIGR climatological
profiles; b) CLASS1 Brazilian climatological profiles. The retrievals obtained over the Amazon
region suggest that temperature profiles inferred by ICI with the TIGR and CLASS1, and using
only the microwave channels, are close to radiossonde “ground truth”. The use of CLASS1
improves the results for moisture profile when compared with TIGR, particularly within low
troposphere.
Atmospheric Profiles
Recent efforts have been concentrated on the development of sensors with high spectral
resolution. The AIRS (Advanced InfraRed Sensor), with 2378 channels, is the first operational
instrument with these characteristics. AIRS sensor was launched onboard EOS-PM (AQUA)
satellite in May 2002, together with microwave units AMSU-A and HSB (Humidity Sensor for
Brazil). The AQUA mission provides a unique opportunity for selecting proper channels in
different atmospheric conditions as well as for improving the “state of the art” in retrieval
techniques. Different selection methods and retrieval techniques are still in development for
information content of AIRS (Goldberg et al. 2003 and 2004, McMillin 2004). The HSB was
similar to AMSU-B sensor but exhibited improved technical performance (Lambrigsten and
Calheiros 2003). Unfortunately, it is no more in operation since the last quarter of 2002 and the
AQUA sounding system is limited to AIRS and AMSU-A information. Nevertheless, the
enormous flux of AIRS data suggests the possibility of partially recovering HSB information in
clear-sky conditions, through emulation of its spectral behaviour by AIRS channels located in
water vapour absorption bands. Present work explores this possibility, seeking for AIRS
channels (or combinations of them) which provide the best simulation of HSB brightness
temperatures for channels centred at 183 GHz. In addition, it is expected to contribute to a better
understanding of the AIRS information content, which in turn may be useful to improve
retrievals of water vapor profiles.
The datasets used in this work correspond to period of the Dry-to-Wet LBA experiment
(August 31 to October 31, 2002) that include: 1) ATOVS/NOAA-16 data; 2) AIRS-AMSUHSB/AQUA data; 3) radiossondes and 4) ICI retrievals over three LBA sites (Guajará-Mirim,
Porto Velho e Ouro Preto). It is important to note that eventual substitution of HSB by AIRS
data is possible only for cloud-free pixels, since AIRS channels are located at infrared spectrum.
Consequently, a previous cloud masking process was applied before starting AIRS channel
selection. The Figure 1 shows the localization of three validation sites over the Amazon region.
Fig. 1 – Validation sites of the LBA campaign.
The starting point for analysis was the detection of AIRS channels highly correlated
with those of HSB. Linear correlation coefficient was assessed for radiances of each AIRS
channel, related to each one of the four HSB channels. A number of Np = 3000 cloud-free pixels
provided the cases for correlation assessment. When ordering AIRS channels by their highervalued correlation with an HSB channel, it was observed that water vapor Jacobians of those
channels showed fair similarity with the corresponding HSB channel. The regression model was
built and the coefficients were assessed for the training day August 31, 2002 and the Figure 2
shows the 3000 cloud-free pixels used. The results were used for simulation of HSB channels
during the whole time interval and comparison with actual (observed) HSB data. Finally, the
simulated HSB data together with AMSU-A/AQUA information were used to recover vertical
profiles of temperature and moisture, using ICI with two first-guess data bases: a) TIGR
climatological profiles (Lavanant et al., 1997); b) CLASS1 Brazilian climatological profiles
(Macedo, 2003).
Fig 2 - The 3000 cloud-free pixels used for the training day August 31, 2002.
Results
The operational ICI-3 in dynamic mode (first guess NWP/CPTEC data) presented better
performance because it used all ATOVS data. However, the retrievals obtained over the
Amazon region (Figure 3) suggest that temperature profiles inferred by ICI with the TIGR and
CLASS1, and using only the microwave channels, are close to radiossonde “ground truth”. The
use of CLASS1 improves the results for moisture profile when compared with TIGR,
particularly within low troposphere. It is important to note that there are differences between
NOAA-16, AQUA and the radiossondes schedules. This fact could influence on the quality of
the results, mainly near the surface.
Conclusion
The AMSU + HSB (simulated) AQUA data yield proper profiles for temperature, in the
“static” ICI mode of operation. The use of CLASS1 improves the results for moisture profile
when compared with TIGR, particularly within low troposphere. Moreover, it will be
appropriated to include typical profiles of the Amazonia occidental region in CLASS1 dataset to
obtain still better results.
Fig. 3 - Temperature (left) and Mixing ratio (right) profiles over three LBA sites:
Guajará-Mirim, Porto velho and Ouro Preto d’Oeste.
Acknowledgments
The authors are thankful to Brazilian CNPq and CAPES for financial support, as well as
to Divisão de Satélites e Sistemas Ambientais, at Centro de Previsão de Tempo e Estudos
Climáticos (DSA/CPTEC), and the Atmospheric Spectroscopy Laboratory (ASL) at University
of Maryland Baltimore County (UMBC) for satellite data and computing facilities.
References
Goldberg, M.D., Qu, Y., MCmillin, L.M., Wolf, W., Zhou, L. and Divakarla, M. 2003. AIRS
near-real-time products and algorithms in support of numerical weather prediction. IEEE
Transactions on Geoscience and Remote Sensing, 2, 379-389.
Goldberg, M.D., Zhu, L., Wolf, W. 2004. Applications of Principal Component Analysis (PCA)
to AIRS data. Proceedings of the 13th Conference on Satellite Meteorology and
Oceanography, Norfolk, USA, 20-23 September, edited by American Meteorological
Society, paper 1.15.
Lambrigtsen, B.H. and Calheiros, R. V. 2003. The humidity sounder for Brazil – An
international partnership. IEEE Transactions on Geoscience and Remote Sensing, 2, 352361.
Lavanant, L.; Brunel, P.; Rochard, G.; Labrot, T.; Pochic, D. 1997. Current Status for the ICI
Retrieval Scheme. In: International TOVS Study Conference, 9. (ITSC), 20-26 February,
Igls, Austria. Proceedings of the Ninth International TOVS Study Conference, Igls: ITSC.
Macedo, S. R. 2003. Um estudo das sondagens do ICI (Inversion Coupled Imager) mediante o
processamento de perfis atmosféricos brasileiros. 2003. 109 p. Dissertação (Mestrado em
Meteorologia) – Instituto Nacional de Pesquisas Espaciais, São José dos Campos.
McMillin, L.M., Super Channels for AIRS retrievals. 2004. Proceedings of the 13th Conference
on Satellite Meteorology and Oceanography, Norfolk, USA, 20-23 September 2004,
edited by American Meteorological Society, paper 6.6.